--------------------------------------BEGIN--------------------------------------------------
Within panel_main.mat there are 3 separate cell structures. 
---------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------

BCFF_bp_merged_dummy  �  Blue Chip Financial (merged participants & FWD_Guidance/QE/MEP events dummy)
BLFF_dummy            �  Bloomberg Financial Aligned (FWD_Guidance/QE/MEP events dummy)
dummySet              -  Manually aligned FWD_Guidance/QE/MEP events


The dummySet:
-Only ever has zeros appended to it to equal the length of the bluechip survey dates (fixed effects are all hard coded to dates starting in 2001 to current)



---------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------
------------------- The 2 "data" structures have nearly identical format: -------------------
---------------------------------------------------------------------------------------------	
---------------------------------------------------------------------------------------------


---------------------------------------------------------------------------------------------
For the BCFF datastructure:
---------------------------------------------------------------------------------------------
	BCFF_bp.dtnum � the datenum of the bluechip survey (mm/01/yyyy)	
	BCFF_bp.cQ - read as row vectors, the datenums for the quarters forecasted by bluechip (convention used - 03/01 (Q1) 06/01 (Q2) 09/01 (Q3) 12/01 (Q4).
	-These are used to from one row to the next to compute if there has been a change in forecasted quarters
	BCFF_bp.shiftQ � the dummy for whether or not there has been a change in forecasted quarters since last survey
	BCFF_bp.fgQEnone - the dummy for forward guidance with no QE at the time of announcement/event
	BCFF_bp.fgQE1 - the dummy for forward guidance with QE1 at the time of announcement/event
	BCFF_bp.fgQE3 - the dummy for forward guidance with QE3 at the time of announcement/event
	BCFF_bp.otherQEnone - the dummy for other programs (MEP,etc.) with no QE at the time of announcement/event
	BCFF_bp.otherQE12 - the dummy for other programs (MEP,etc.) with QE1 or QE2 at the time of announcement/event
	BCFF_bp.partArray � An ordered list of strings with rows matching the BCFF.fcasters.f_<#> "participant #."  The strings served as unique IDs and were thrown through a merging algorithm to ensure all participants were correctly matched to their forecasts over the 10Yr history.
	BCFFv.partArray_all � Shows the different unique strings that have been matched as the same forecasting participant.  
{i.e.'banc of america securities, llc';
'banc of america securities';
'banc of amercia securities';
'banc of america-merrill lynch';
'bank of america merrill lynch';
'bank of america';
'bank of america-merrill lynch';
 }
	
BCFF_bp.('indicator').f_<#> � the forecast value/s for the indicator from participant <#>
	-there are 6 forecast values in the bluechip reports (6 Quarter horizon), and a single forecasted value for each bloomberg survey indicator (may not want these)
BCFF_bp.('indicator').cons  - the consensus BlueChip forecast (MEAN of participants)	
	



---------------------------------------------------------------------------------------------
For the BLFF datastructure:
---------------------------------------------------------------------------------------------
	BLFF.dtnum � the date of the bluechip survey "release date" [num is always first of the month (mm/01/yyyy)]
	BLFF.'indicator'.dtrelnum - the bloomberg release date for a given indicator, aligned to upcoming bluechip "release date"
	
	BLFF.dtcol � the date of the bluechip survey "last collection date" [col is somewhere between (mm/19-26/yyyy)]
	BLFF.'indicator'.dtrelcol - the bloomberg release date for a given indicator, aligned to upcoming bluechip "last collection date"


	BLFF.'indicator'.dtrelnum - the bloomberg "indicator release date" with Blue Chip "release date" alignment
	BLFF.'indicator'.forecastnum - the bloomberg indicator forecast with "release date" alignment
	BLFF.'indicator'.actualnum - the bloomberg forecast with "release date" alignment

	BLFF.'indicator'.dtrelcol - the bloomberg "indicator release date" with Blue Chip "last collection date" alignment
	BLFF.'indicator'.forecastcol - the bloomberg indicator forecast with "last collection date" alignment
	BLFF.'indicator'.actualcol - the bloomberg forecast with "last collection date" alignment
	
"2-4"  - The additional news (survey-actual) that occurs between any BlueChip window, but precedes the last news available in that window (Weekly data, the dtrelnum will correspond to the weekly news available  just before the next BlueChip survey, and the dtrelnum2 corresponds to the weekly news available 2 weeks prior to the next Bluechip survey.

	BLFF.'indicator'.dtrelnum2 
	BLFF.'indicator'.forecastnum2 
	BLFF.'indicator'.actualnum2 

	BLFF.'indicator'.dtrelcol2 
	BLFF.'indicator'.forecastcol2 
	BLFF.'indicator'.actualcol2 



	For daily, end of day data, we take the inter-survey window to run from the begining of the previous survey to the end of the current survey.
	(i.e. prev survey == '03/22-23/2015', current survey == '4/25-26/2015', the window would be 3/22-4/26/2015) -- This window of values is aligned to show the conditional data up to a release (dtcol would appear as 4/26/2015 in the hypothetical situation just used)
	NaN's will appear at the start of all these vectors of logchange, because prior to the 1st daily data available there is often no prior survey date to construct a return window (and do not construct an artificial 1 month/30 day window)
	BLFF.'indicator'.actualcolWindow - change in the inter-survey bluechip collection dates
	BLFF.'indicator'.actualcolEvents - 2 day - change over the FOMC/other events in the sample
	BLFF.'indicator'.actualcolWsubE - change in the inter-survey bluechip collection dates, leaving out the contribution of the 2 day event
	
	* note: if there is more than 1 event in a particular survey window, the difference is summed together (this results in events for 16/170 survey windows.  There are 20 total events).  
	* If a 2day event were to take place in the last day of the survey, only the 1 day falling inside the survey window would be counted in the net event difference in the period.  This does not appear to occur.

---------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------
General notes:
---------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------
*Pulled bloomberg actual values back to 2000.
*Have bluechip survey results back to 2001.
*The BCFF and BLFF dtnum goes from 01/2001-03/2015

---------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------
Transformations:
---------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------
basicdiff = {'FNMGVN5','FNMGVN10','FNMGVN20','FNMGVN30',...
    'MoodysAAA10','MoodysBAA10','CFSI','CESIUSD','MDSICSUS','ECSURPUS',...
    'ECSUSUUS','ECSULBUS','ECSUHOUS','ECSURTUS','ECSUINUS','ECSUPEUS',...
    'SEPUI'}; % the FNMGVN# have already been made basis point

bpdiff = {'GNMA15','GNMA30','FMNMA15','FMNMA30','GOLD15','GOLD30',...
    'MBS15','MBS30','BE1','BE5','BE10','BE20','BE30','DGS1','DGS5','DGS10',...
    'DGS20','DGS30','TIPS10','TIPS30'};

lnchng = {'BBOX','SP500','DJIA','FX_EDUS','CDS5','CDS7','VIX','SEPUIN',...
    'RASC','RASI','SEMUI'};

* Some of the series (ECS*, SEPUI, CFSI) have been constructed as a mean zero index, so we move them to the basic difference transformation (log(0) = inf)







------------------------------------------END------------------------------------------------
